Dataset utilities

Release:

6.1.2

Date:

Oct 28, 2026

cached_versions()	Return the dataset versions present in the cache.
cache_root()	Return the effective cache root.
categories()	Returns the dataset categories mapped to their graph names.
clear_cache([keep])	Remove the cached dataset versions, optionally keeping one.
download_graph(name)	Deprecated alias of graph_dir().
download_reachable_pairs()	Download the versioned reachable-pairs CSV from the dataset storage.
graph_dir(name[, verify])	Return the local directory of a graph, downloading it on a cache miss.
graph_info(name)	Returns the registry record of one graph.
graph_names()	Returns the names of all downloadable graphs, sorted.
reachable_pairs([graph, grammar, category, ...])	Return reference reachable-pair counts.
version_dir([version])	Return the cache directory of a dataset version.

class GraphInfo(name: str, category: str, num_nodes: int, num_edges: int, size_mb: float, sha256: str, queries: tuple[QueryInfo, ...])

The registry record of one graph.

Attributes:

namestr

The graph name (the archive is <name>.tar.gz).

categorystr

The dataset category the graph belongs to.

num_nodesint

The number of vertices (the declared MTX dimension).

num_edgesint

The number of stored edges (the sum of the MTX nnz).

size_mbfloat

The archive size in MB (three decimals).

sha256str

The sha256 hex digest of the <name>.tar.gz archive.

queriestuple[QueryInfo, …]

The queries shipped with the graph, sorted by (class, name).

class QueryInfo(query_class: str, name: str, representations: tuple[str, ...])

One query of a graph (a queries/ dir of its archive).

Attributes:

query_classstr

The query class (cfpq, rpq, or mcfpq).

namestr

The query name (the directory name under queries/<class>/).

representationstuple[str, …]

The representation formats present for the query (e.g. ("cnf", "rsm")), sorted.